How can we use semantic segmentation for handwritten character recognition?
Dear Vrushali Lanjewar ,
Segmentation is nothing but breaking the whole image into subparts to process them further.
Segmentation of image is done in the following sequence : Line level Segmentation Word level Segmentation Character level Segmentation
https://towardsdatascience.com/segmentation-in-ocr-10de176cf373
Regards,
Shafagat
Semantic segmentation is a technique used in image processing and computer vision to identify and classify different objects in an image. It is often used for image recognition tasks, such as character recognition. To use semantic segmentation for character recognition, you would first need to train a semantic segmentation model on a dataset of labeled images that include characters. This would involve feeding the model a large number of images, along with the corresponding labels for each image, so that it can learn to identify and classify different characters.
Once the model is trained, you can then use it to segment characters in new images. This involves providing the model with an input image and asking it to predict the labels for each pixel in the image. The model will then output a segmented image, with each pixel labeled according to the character it belongs to. This can be used to identify and recognize the different characters in the image.
One important thing to note is that semantic segmentation is not a perfect solution for character recognition, and it may not always be able to accurately identify and classify every character in an image. However, it can be a useful tool in certain situations, particularly when combined with other techniques, such as optical character recognition (OCR).
@Sant Gadge Baba Amravati University
Vrushali Lanjewar ,
I would take a look at[1] where they use semantic segmentation in their work of text detection. I would see the role of semantic segmentation more as a substitute for the skew detection algorithms used in OCR.
[1]TextFuseNet: Scene Text Detection with Richer Fused Features by Ian Ye , Zhe Chen, Juhua Liu and Bo Du
https://www.ijcai.org/proceedings/2020/0072.pdf
- Badges
- Science topic
- Similar topics
- Filing